To most of you, I suspect this is nothing new. I'm not what you'd call a highly experienced anime watcher, I have only extremely rarely tried following currently airing seasonal anime schedules,...
Exemplary
To most of you, I suspect this is nothing new. I'm not what you'd call a highly experienced anime watcher, I have only extremely rarely tried following currently airing seasonal anime schedules, for the most part I watch older stuff based on acclaim and personal appeal, and I have maybe subscribed to Crunchyroll for a lifetime cumulative total of 2 months, probably almost a decade ago.
I didn't really know this was going on on Crunchyroll but I saw this article elsewhere and found it absolutely captivating. The author goes into lots of detail about the behind the scenes industry side of this, the technical details and software of typesetting subtitles, the history of Crunchyroll and Funimation's subtitling and typesetting practices, and ultimately demonstrates how and why they're getting worse. The article itself, almost as if it demonstrates the same values, is presented and formatted in a really pleasing way, too.
My first run-in with really noticing how big of a difference really nice typesetting in subtitles makes was the Chyuu-PAS fansub of Revue Starlight, which I watched. I was impressed with the choice to use these art deco style fonts at a time when I was not even really aware of the difference between official subs and fansubs. Then I wanted to show stuff from the show to my friends, (such as the transformation sequence) which was what I had seen clips from and got me interested in the show. If I've caught your interest in this sequence by mentioning it, you can watch a version of it here.
Originally it was officially on HiDive, and I know it did not do song lyrics (likely similar separate-licensing-for-music reasons as given in the article) but I can't comment on what typesetting it had as I never saw that version. Today, the official service it is on is Prime Video, which due to the constraints of their subtitle system, would not have any of the typesetting seen earlier in my comment. I don't have Prime, so I can't go gathering comparison shots.
Anyway, I'm definitely straying away from the topic a bit, but really the experience of watching this fansub in particular and that if I tried to recommend Revue Starlight to my friends today, knowing they're gonna see it with Prime Video's bottom-text subtitles only is what primed me to passionately and vehemently agree with and want to share this article.
I hope y'all don't mind if I come off naive sharing this here, maybe none of this is anything new to those of you who hang around and post here in ~anime but I just hope that maybe it sparks a bit of discussion, or maybe for those of you already aligned with this, perhaps by sharing the article I've offered you a conveniently linkable version of a well researched and formatted argument against Crunchyroll's new practices.
Hey, I read this blog/article yesterday and it's really great. I just wanted to say you shouldn't be worried about coming off as naive for sharing this. The author, Daiz, is a prominent figure in...
Hey, I read this blog/article yesterday and it's really great. I just wanted to say you shouldn't be worried about coming off as naive for sharing this. The author, Daiz, is a prominent figure in the scene. They have been around for a long time in both fan and official capacities pushing for various improvements. This is niche knowledge covered by an expert. It provides a fantastic overview of the situation in a very accessible manner. While some portions of it are "obvious" to people who have been involved in the anime (and specifically fansubbing) scene for a while, much of it is newly confirmed information on the specific workflows of companies and how that has lead to the current issue being addressed.
I also appreciate the detailed write-up and examples of your own experiences relating to quality subtitles and frustrations with some of those which are available through official channels.
Thanks for the reassurance haha. There might be an element of pre-excusing myself if the thing I posted didn't land with that aspect of my comment. It just happens to be my first foray into ~anime...
Thanks for the reassurance haha. There might be an element of pre-excusing myself if the thing I posted didn't land with that aspect of my comment.
It just happens to be my first foray into ~anime on tildes at all and my first time posting a topic on tildes full stop, only been commenting up to this point. It's not a joke in my bio where it says my bookmark leads straight to ~games so I just didn't know what the ~anime section would be when it comes to industry type topics such as this.
It's good to know the article confirms things that those already in the know more generally already expected. I never heard of Daiz until this article but like I said before that'll be a consequence of my very casual, community-less anime viewing habits. I'm not in the know, so I don't know what the ~anime community at Tildes knows or the broader online anime community's disposition on matters like these either, so that's what I mean by naive, I'm just posting with no understanding of whether this is already a prevailing topic or whether this is new or old to the average reader around here.
As someone who studied media translation (subtitling included), anime is really its own beast that you need separate specifications and guidelines for due to the vast amount of on-screen text as...
As someone who studied media translation (subtitling included), anime is really its own beast that you need separate specifications and guidelines for due to the vast amount of on-screen text as well as the different alphabet that the majority of your audience will not be able to read. It's an absolute shame that Crunchyroll and some other companies do not take this seriously and end up diminishing their paid viewer experience for no good reason (I mean I know the reason is profit with the least amount of work possible, I just don't consider that a good reason).
Traditionally, when subtitling, you used to have a two line limit at the bottom of the screen for most situations (and a character limit per second, usually about 42-48), and then you may occasionally use the top of the screen for the on-screen text, if that text is necessary to comprehend the narrative (otherwise you'd ignore it so as to not clutter the screen). This was mainly a result of technical limitations of traditional formats like TV and DVD players, where fancy styling was not available so you couldn't encode that information in your subtitles. It was also important for subtitles to be modifiable for accessibility reasons as well, so the simpler the better, and in a traditional setting a subtitle file contains little more than timestamps and the text itself.
But for anime, especially nowadays, there are always so many scenes with text on screen (just think of how many fantasy rpg anime with menu screens there are these days) that it should be imperative for companies like Crunchyroll to have guidelines for them and not ignore them completely. Web players are also capable of encoding all types of styling as well, and if accessibility is an issue, you can always include an non stylized option for the people who need it, because modern players allow for such implementations, unlike TV in the old days.
There's also audience expectation. Anime fans are much more inclined to be interested in everything that's on screen because they're already conscious that they're watching a work from another culture and that difference is a huge part of the appeal, so losing out on that vital information diminishes the experience.
Like, I get it. As a translator I get that deadlines are tight and sometimes technical limitations don't allow you to do everything in the most pristine way, but we have ample evidence that something better than what is currently being provided is possible, and Crunchyroll and other anime licensing companies should absolutely be held to that standard if they are charging a fee for their service.
Yeah, this has been upsetting me for a little while now. I already almost refuse to watch Netflix and Amazon anime because it's really not uncommon to just have them do some janky subtitles at...
Yeah, this has been upsetting me for a little while now. I already almost refuse to watch Netflix and Amazon anime because it's really not uncommon to just have them do some janky subtitles at some point that end up just detracting from the experience. Sure, there may not be things like on screen text or multiple people talking every episode, and sometimes don't happen in a show at all, but there's no way for me to know that without watching it. And if I'm watching it, then when those things do crop up, I'm pulled out of the show as I suddenly need to decipher who's saying what and what signs are where.
In the event that a fansub group doesn't pick a Netflix/Amazon show up in the first couple weeks, I often don't bother. I'd much rather just wait a few years, maybe remember that they had a show I wanted to watch, then do a quick check to see if anyone has put out some quality fansubs.
But, alas, at least for me, this enshittification may be coming at an opportune time; my interest in anime has already been rapidly dwindling these past couple years, this season definitely included. So it's possible my dwindling interest may be due in part to this, but it's hard to say. Maybe I'm just being a drama queen, but if this trend keeps continuing, then I don't know how much longer I can watch anime.
I learned about this a few weeks back through a YouTube video. Quite an unfortunate degradation in behavior. I do get why they're doing it, but for such a huge company I personally find their...
I learned about this a few weeks back through a YouTube video. Quite an unfortunate degradation in behavior.
I do get why they're doing it, but for such a huge company I personally find their reasoning a bit flimsy. Sony has far more than the resources required to build an in-house tool that properly does everything they want.
And on soft vs hard subs, it's unfortunate how many devices don't bother to support PGS. My TV box is a super cheap like $20 thing and I'm pretty sure it does, but Rokus and such don't. If it was more widely supported they could use PGS to soft sub while still making it look almost as good (it's possible to hard sub animating subtitles though I don't think you can do that in PGS, but that kind of thing is exceptionally rare). A company the size of Sony could potentially have the leverage to get more support added, at least as long as they're pushing a standard format that those companies may find it reasonable to support.
For anyone unfamiliar with PGS, it's an image based subtitle format. This isn't exactly how its done, but this is close enough to explain it for the purpose of this discussion: imagine the subtitles are actually just a collection of pngs alongside a data stream that contains instructions of when and where to overlay each image. They're obviously then more annoying to edit, but no one says they have to edit them as PGS. They could edit them in a tool designed for their workflows that just has a PGS export button.
The trick here is what's effectively on-the-fly hardsubbing, and it removes any endpoint device dependency issues from the equation. Even a free home streaming box app like Plex has the ability to...
If it was more widely supported they could use PGS to soft sub while still making it look almost as good (it's possible to hard sub animating subtitles though I don't think you can do that in PGS, but that kind of thing is exceptionally rare).
The trick here is what's effectively on-the-fly hardsubbing, and it removes any endpoint device dependency issues from the equation. Even a free home streaming box app like Plex has the ability to stream subtitles by combining them with the video as it's sent out.
That's a usage of transcoding and can be rather computationally expensive, especially to perform at scale. I use a Raspberry Pi as a media server and it works great for direct play (ie. just...
That's a usage of transcoding and can be rather computationally expensive, especially to perform at scale. I use a Raspberry Pi as a media server and it works great for direct play (ie. just forwarding the bits from storage to network) but it can't really handle on-demand transcoding. It not being able to handle the PGS from my Blu-rays is why I even care about what TV boxes support PGS. Using a device that doesn't really support on-demand transcoding was a semi-intentional decision on my part because I didn't like hearing a computer fan spin up when I started watching things because to me the fan kicking on means the machine is working harder than I want it to.
It can get better if you have hardware that handles transcoding so you don't need to use the slower software solutions, but a service provider isn't going to want to build out a fleet of such devices unless they have to, especially when they're also quite a bit more expensive to operate. That, combined with a bunch of other factors related to mediums where you don't have that option (ex. broadcast, DVDs, etc.), mean that it is generally more sane to have the user device do the conversions.
User devices can also generally perform said conversions far more efficiently. This is not only because the device will generally have hardware specifically designed for video compositing, but all because rendering a frame and then overwriting some pixels is the easy part. Your TV will do that basically any time you open a menu. The hard part is encoding that new stream of frames into an equivalently high quality video stream because video encoding is computationally complex as it tries to optimize for both size and quality. This usually is at the cost of performance since you can generally just store and serve the result. It's basically the "pick 2: good, fast, cheap" thing renamed to "quality, size, encoding time/cost".
So all that's to say that on-demand hard subbing would be expensive to pay for the compute power for if you were a large operator. It would also waste a lot of electricity and likely bandwidth (on-demand transcoding usually has to favor fast even if it could compress it better with more compute time). It's more reasonable to do what Crunchyroll used to do: proactively burn in subs for every language stored as separate video files, but that gives file management woes and only really works when the service is designed to work like that.
If a major streamer can't figure out their caching infrastructure and such to balance the burdens, maybe they need to work on their app so it can handle video overlays on a basic level.
That's a usage of transcoding and can be rather computationally expensive, especially to perform at scale
If a major streamer can't figure out their caching infrastructure and such to balance the burdens, maybe they need to work on their app so it can handle video overlays on a basic level.
I use Jellyfin rather than Plex, but I know Jellyfin decides whether to direct play or transcode on the server based on the player hardware. This is because many pieces of video playing hardware...
I use Jellyfin rather than Plex, but I know Jellyfin decides whether to direct play or transcode on the server based on the player hardware. This is because many pieces of video playing hardware have very weak general purpose compute and use dedicated hardware for decoding and compositing features. This is the same reason that some formats also trigger transcoding with no overlay (if memory serves Rokus less than 4k don't support HEVC). Ultimately this leaves them designing for lowest common denominator devices and very limited ability to implement interesting features in app. Plex and Jellyfin and such can get away with it in ways that actual services can't because your library being broken is your own fault and problem. I have wondered if it would be possible to implement PGS as a UI-like overlay, but presumably there's a reason no one does.
As for what that means they should do though is messy. To me it means that they should continue to do what they used to of having a copy of the video per language and with burned in subtitles on each. But, as said, it kind of requires that the infrastructure has that very abnormal design. Part of the problem with that design is that you can't as cleanly license the content to be playable on players without that design, such as how historically the Crunchyroll content looked good on Crunchyroll but bad on Amazon. The approach of having it be bad everywhere is a pretty bad solution though. So part of why Sony wants this change is that Crunchyroll has been producing a barely-compatible library for licensing.
So there's a whole mess of conflicting stuff. Crunchyroll should have an app that works in the ideal way, but they also should have a library that works in the ideal way on the Crunchyroll Amazon Prime channel that they don't have control over the code for. They already did effectively the exact thing you were suggesting except that they did it up front rather than on-demand, but it gave them other problems. Specifically they were doing the work twice in different ways for the different distribution channels with some of those channels having objectively superior results.
These are the reasons that my original comment was about using a more flexible standard. Leveraging a more powerful standard is something they could presumably convince other distribution channels to do since it's a net benefit for their apps where "rebuild your infrastructure to work in this weird way ours does" wouldn't realistically ever go anywhere.
Yes, yes, pedantry aside, basic video compositing is something a major service like Crunchyroll needs to help partners get going in return for the non-trivial cut of CR's money they're skimming.
Yes, yes, pedantry aside, basic video compositing is something a major service like Crunchyroll needs to help partners get going in return for the non-trivial cut of CR's money they're skimming.
This was a really good read. I know a lot more about subtitling than I did before. I am primarily a dub watcher, but the on screen text subtitles are crucial. It's definitely making me reconsider...
This was a really good read. I know a lot more about subtitling than I did before. I am primarily a dub watcher, but the on screen text subtitles are crucial. It's definitely making me reconsider my subscription.
I'm hoping licenses start going to other services. Just to add to competition to possibly improve subtitling. That or a resurgence of fan subtitling. I know it's not easy, but fans on the internet will churn it out quickly. At least based on my experience following OPM manga
I would not count on competition nor fans to fix this issue. With competition, it's not about being better than your competition, it's about not being worse. If you hold the rights to a show,...
I would not count on competition nor fans to fix this issue. With competition, it's not about being better than your competition, it's about not being worse. If you hold the rights to a show, people will be forced to come to you to watch it, regardless of your quality. So as long as it's not unreadable, people will watch anything sufficiently popular. More than that, I find majority of people don't even care. Often when I talk about anime with people, they talk about things they watched on Netflix. Stuff that I often don't bother with because of their useless subtitling.
One more extreme example is Cyberpunk: Edgerunners. When it aired, my friends were raving about it, but I was sitting around waiting for it to get some fansubs. After a while, I eventually caved and, to no one's surprise, had some major issues in just the first episode. See, at several points, people get phone calls on their AR phones (or whatever they are), and the artists then animate a scrolling transcription of what the other side is saying. So Netflix, in their infinite wisdom, just decided to not put subtitles for that side of the conversation and let us rely on the artists' translation.
However, the animated phone text (if I remember correctly) is some sort of neon green in a really janky font and slightly green background, making it difficult to read at a glance. Worse than that, there were even one or two scenes where it showed the person receiving the call from the front. So the text being a holographic display in front of them, was then mirrored and completely unreadable. As I was watching with the Japanese voice-over and English subtitles, I had to rewind and pause several times just to get the gist of what they were saying. When I vented about this to my friends, they just said, "Did it? Oh, yeah. After your specific example, I guess they did. Well, they do it properly in the rest of the show." They did not.
As for fansubs, I can't help but feel as though manga and anime are just completely different beasts when it comes to translations. Some of my assumptions on it may be a little off as I don't read manga pretty much at all, so keep that in mind. The first, and probably most notable, aspect is that anime already gets same day translations. With manga, I don't believe this is the case at all. So there's an incentive to beat the official translations; you get a lot more prestige, or ad revenue, or whatever it is the translators are chasing by simply staying ahead of the official translation.
The next is the work involved. Anime often covers several chapters of manga in a single episode. Chapters that are released weekly, fortnightly, monthly, or sometimes even less frequently than that. So there's often just more to translate. Granted, with a lot of them, they can rely on the source material that they are likely already familiar with, so they may be ahead of things a tiny bit in a way. But also, the work to make sure things are readable. With manga (and again, this might be my assumptions at play), there's not as much a worry in things being perfectly legible¹ because people can read it at their own pace. With anime, there's very much a limit to what you can do. There's limits on words per minute. When multiple people are talking at the same time, you can't just put each person's text into their respective bubbles, you have to figure out how to convey who's talking. When there's a sign, it needs to be readable at a glance without distracting the viewer away from the things that people are saying. And lastly, the timing; the text needs to line up with certain things to minimize the distraction that the changing subtitles may cause.
And lastly, fansubs have already been dying out. Most the new fansubs I see pop up are things from before Crunchyroll became such a big player. Other times, they're just restylized, upscaled, or fixed versions of a fansub that someone else released. Other times, they're just Crunchyroll/Netflix/whoever's official subs with proper typesetting and signing. I've little doubt the amount of groups that are doing original subtitles for currently airing shows at any given time are countable on one hand, and then they're generally at least a week behind the official subs. Good luck avoiding spoilers if you want to discuss that show.
So, I really don't believe there's any hope that things will improve without Crunchyroll or one of the other major corporations actively deciding to do better. We could theoretically push them towards improving by unsubscribing until they do. But as I said, I don't believe majority of people care enough. So it would be a slow burn and by the time they take measures to improve, the writing for them may already be on the wall.
¹I have no doubt manga translators have some standards and put a lot of work into making sure it's all readable and fits with the art. However, I do believe they occasionally use a less legible font at times in order to convey certain emphasis or tone. Something the original Japanese version is likely doing as well.
You're mostly right. Only some big series like One Piece have simultaneous international releases. Though for those big series, there's a thriving market for acquiring early (usually Chinese)...
The first, and probably most notable, aspect is that anime already gets same day translations. With manga, I don't believe this is the case at all. So there's an incentive to beat the official translations; you get a lot more prestige, or ad revenue, or whatever it is the translators are chasing by simply staying ahead of the official translation.
You're mostly right. Only some big series like One Piece have simultaneous international releases. Though for those big series, there's a thriving market for acquiring early (usually Chinese) copies then rapidly translating them before the official release.
Most the new fansubs I see pop up are things from before Crunchyroll became such a big player. Other times, they're just restylized, upscaled, or fixed versions of a fansub that someone else released. Other times, they're just Crunchyroll/Netflix/whoever's official subs with proper typesetting and signing.
Yep, it's kind of disappointing, but also the signs are the thing that Netflix et al are most likely to fuck up. Even for dubs, it can be worthwhile to find a good release with translated signs. It's shockingly common for the dub to have no sign translations if a show is less popular...
¹I have no doubt manga translators have some standards and put a lot of work into making sure it's all readable and fits with the art. However, I do believe they occasionally use a less legible font at times in order to convey certain emphasis or tone. Something the original Japanese version is likely doing as well.
I've worked as a cleaner for manga fanslations, so I can answer this! It's very very involved if you go all-in on translating everything. The biggest example is sound effects. The Japanese love their sound effects, and it takes a lot of time to carefully remove and replace them. Sadly, I don't have any typeset examples on hand, but I have a few screenshots of a couple panels I cleaned. I was a complete novice when I did these two panels, so you can find some inconsistent gradients and patterns if you zoom in.
Ignoring sound effects (as many official releases do), cleaning is still required for panels where the words are directly written on the art. If they're not in a speech bubble, it takes time and skill to properly erase and replace.
Poor subtitling is actually a big reason I've never considered getting a Crunchyroll or Funimation plan. I'm generally not watching the newest, latest shows, so by the time I'm convinced that...
Poor subtitling is actually a big reason I've never considered getting a Crunchyroll or Funimation plan. I'm generally not watching the newest, latest shows, so by the time I'm convinced that something's worth watching, way better fansubs have been released than what you get through the official channels. Sad to say, but unless you're invested in being part of all the online hype every week (or watch anime primarily dubbed, I suppose), I don't see the experience on CR/Amazon/Netflix as better than sailing the high seas.
The irony is that while you used to maybe have to be skeptical about the quality of fansub translations back in the day, now the fansubs are based on, and improve, the official subs.
To most of you, I suspect this is nothing new. I'm not what you'd call a highly experienced anime watcher, I have only extremely rarely tried following currently airing seasonal anime schedules, for the most part I watch older stuff based on acclaim and personal appeal, and I have maybe subscribed to Crunchyroll for a lifetime cumulative total of 2 months, probably almost a decade ago.
I didn't really know this was going on on Crunchyroll but I saw this article elsewhere and found it absolutely captivating. The author goes into lots of detail about the behind the scenes industry side of this, the technical details and software of typesetting subtitles, the history of Crunchyroll and Funimation's subtitling and typesetting practices, and ultimately demonstrates how and why they're getting worse. The article itself, almost as if it demonstrates the same values, is presented and formatted in a really pleasing way, too.
My first run-in with really noticing how big of a difference really nice typesetting in subtitles makes was the Chyuu-PAS fansub of Revue Starlight, which I watched. I was impressed with the choice to use these art deco style fonts at a time when I was not even really aware of the difference between official subs and fansubs. Then I wanted to show stuff from the show to my friends, (such as the transformation sequence) which was what I had seen clips from and got me interested in the show. If I've caught your interest in this sequence by mentioning it, you can watch a version of it here.
While looking for clips to share in turn, I noticed how much worse other versions were. (seriously, what the hell is this?) (I'm not sure where those came from or what the official ones on HiDive looked like) I was so disappointed. I mean, Chyuu-PAS did it so well! Look at the name on the chalk board, it looks like chalk and is appropriately matching the camera's perspective of the chalkboard. It's even in the long-shot of the classroom while not being practically readable. It's almost believable that the original animators put it in there, like she wrote her name on the blackboard in romaji also, except that it disappears when you turn the subs off.
Originally it was officially on HiDive, and I know it did not do song lyrics (likely similar separate-licensing-for-music reasons as given in the article) but I can't comment on what typesetting it had as I never saw that version. Today, the official service it is on is Prime Video, which due to the constraints of their subtitle system, would not have any of the typesetting seen earlier in my comment. I don't have Prime, so I can't go gathering comparison shots.
Anyway, I'm definitely straying away from the topic a bit, but really the experience of watching this fansub in particular and that if I tried to recommend Revue Starlight to my friends today, knowing they're gonna see it with Prime Video's bottom-text subtitles only is what primed me to passionately and vehemently agree with and want to share this article.
I hope y'all don't mind if I come off naive sharing this here, maybe none of this is anything new to those of you who hang around and post here in ~anime but I just hope that maybe it sparks a bit of discussion, or maybe for those of you already aligned with this, perhaps by sharing the article I've offered you a conveniently linkable version of a well researched and formatted argument against Crunchyroll's new practices.
Hey, I read this blog/article yesterday and it's really great. I just wanted to say you shouldn't be worried about coming off as naive for sharing this. The author, Daiz, is a prominent figure in the scene. They have been around for a long time in both fan and official capacities pushing for various improvements. This is niche knowledge covered by an expert. It provides a fantastic overview of the situation in a very accessible manner. While some portions of it are "obvious" to people who have been involved in the anime (and specifically fansubbing) scene for a while, much of it is newly confirmed information on the specific workflows of companies and how that has lead to the current issue being addressed.
I also appreciate the detailed write-up and examples of your own experiences relating to quality subtitles and frustrations with some of those which are available through official channels.
Also Revue Starlight kicks ass.
Thanks for the reassurance haha. There might be an element of pre-excusing myself if the thing I posted didn't land with that aspect of my comment.
It just happens to be my first foray into ~anime on tildes at all and my first time posting a topic on tildes full stop, only been commenting up to this point. It's not a joke in my bio where it says my bookmark leads straight to ~games so I just didn't know what the ~anime section would be when it comes to industry type topics such as this.
It's good to know the article confirms things that those already in the know more generally already expected. I never heard of Daiz until this article but like I said before that'll be a consequence of my very casual, community-less anime viewing habits. I'm not in the know, so I don't know what the ~anime community at Tildes knows or the broader online anime community's disposition on matters like these either, so that's what I mean by naive, I'm just posting with no understanding of whether this is already a prevailing topic or whether this is new or old to the average reader around here.
Thanks again for your kind reply!
As someone who studied media translation (subtitling included), anime is really its own beast that you need separate specifications and guidelines for due to the vast amount of on-screen text as well as the different alphabet that the majority of your audience will not be able to read. It's an absolute shame that Crunchyroll and some other companies do not take this seriously and end up diminishing their paid viewer experience for no good reason (I mean I know the reason is profit with the least amount of work possible, I just don't consider that a good reason).
Traditionally, when subtitling, you used to have a two line limit at the bottom of the screen for most situations (and a character limit per second, usually about 42-48), and then you may occasionally use the top of the screen for the on-screen text, if that text is necessary to comprehend the narrative (otherwise you'd ignore it so as to not clutter the screen). This was mainly a result of technical limitations of traditional formats like TV and DVD players, where fancy styling was not available so you couldn't encode that information in your subtitles. It was also important for subtitles to be modifiable for accessibility reasons as well, so the simpler the better, and in a traditional setting a subtitle file contains little more than timestamps and the text itself.
But for anime, especially nowadays, there are always so many scenes with text on screen (just think of how many fantasy rpg anime with menu screens there are these days) that it should be imperative for companies like Crunchyroll to have guidelines for them and not ignore them completely. Web players are also capable of encoding all types of styling as well, and if accessibility is an issue, you can always include an non stylized option for the people who need it, because modern players allow for such implementations, unlike TV in the old days.
There's also audience expectation. Anime fans are much more inclined to be interested in everything that's on screen because they're already conscious that they're watching a work from another culture and that difference is a huge part of the appeal, so losing out on that vital information diminishes the experience.
Like, I get it. As a translator I get that deadlines are tight and sometimes technical limitations don't allow you to do everything in the most pristine way, but we have ample evidence that something better than what is currently being provided is possible, and Crunchyroll and other anime licensing companies should absolutely be held to that standard if they are charging a fee for their service.
Yeah, this has been upsetting me for a little while now. I already almost refuse to watch Netflix and Amazon anime because it's really not uncommon to just have them do some janky subtitles at some point that end up just detracting from the experience. Sure, there may not be things like on screen text or multiple people talking every episode, and sometimes don't happen in a show at all, but there's no way for me to know that without watching it. And if I'm watching it, then when those things do crop up, I'm pulled out of the show as I suddenly need to decipher who's saying what and what signs are where.
In the event that a fansub group doesn't pick a Netflix/Amazon show up in the first couple weeks, I often don't bother. I'd much rather just wait a few years, maybe remember that they had a show I wanted to watch, then do a quick check to see if anyone has put out some quality fansubs.
But, alas, at least for me, this enshittification may be coming at an opportune time; my interest in anime has already been rapidly dwindling these past couple years, this season definitely included. So it's possible my dwindling interest may be due in part to this, but it's hard to say. Maybe I'm just being a drama queen, but if this trend keeps continuing, then I don't know how much longer I can watch anime.
I learned about this a few weeks back through a YouTube video. Quite an unfortunate degradation in behavior.
I do get why they're doing it, but for such a huge company I personally find their reasoning a bit flimsy. Sony has far more than the resources required to build an in-house tool that properly does everything they want.
And on soft vs hard subs, it's unfortunate how many devices don't bother to support PGS. My TV box is a super cheap like $20 thing and I'm pretty sure it does, but Rokus and such don't. If it was more widely supported they could use PGS to soft sub while still making it look almost as good (it's possible to hard sub animating subtitles though I don't think you can do that in PGS, but that kind of thing is exceptionally rare). A company the size of Sony could potentially have the leverage to get more support added, at least as long as they're pushing a standard format that those companies may find it reasonable to support.
For anyone unfamiliar with PGS, it's an image based subtitle format. This isn't exactly how its done, but this is close enough to explain it for the purpose of this discussion: imagine the subtitles are actually just a collection of pngs alongside a data stream that contains instructions of when and where to overlay each image. They're obviously then more annoying to edit, but no one says they have to edit them as PGS. They could edit them in a tool designed for their workflows that just has a PGS export button.
The trick here is what's effectively on-the-fly hardsubbing, and it removes any endpoint device dependency issues from the equation. Even a free home streaming box app like Plex has the ability to stream subtitles by combining them with the video as it's sent out.
That's a usage of transcoding and can be rather computationally expensive, especially to perform at scale. I use a Raspberry Pi as a media server and it works great for direct play (ie. just forwarding the bits from storage to network) but it can't really handle on-demand transcoding. It not being able to handle the PGS from my Blu-rays is why I even care about what TV boxes support PGS. Using a device that doesn't really support on-demand transcoding was a semi-intentional decision on my part because I didn't like hearing a computer fan spin up when I started watching things because to me the fan kicking on means the machine is working harder than I want it to.
It can get better if you have hardware that handles transcoding so you don't need to use the slower software solutions, but a service provider isn't going to want to build out a fleet of such devices unless they have to, especially when they're also quite a bit more expensive to operate. That, combined with a bunch of other factors related to mediums where you don't have that option (ex. broadcast, DVDs, etc.), mean that it is generally more sane to have the user device do the conversions.
User devices can also generally perform said conversions far more efficiently. This is not only because the device will generally have hardware specifically designed for video compositing, but all because rendering a frame and then overwriting some pixels is the easy part. Your TV will do that basically any time you open a menu. The hard part is encoding that new stream of frames into an equivalently high quality video stream because video encoding is computationally complex as it tries to optimize for both size and quality. This usually is at the cost of performance since you can generally just store and serve the result. It's basically the "pick 2: good, fast, cheap" thing renamed to "quality, size, encoding time/cost".
So all that's to say that on-demand hard subbing would be expensive to pay for the compute power for if you were a large operator. It would also waste a lot of electricity and likely bandwidth (on-demand transcoding usually has to favor fast even if it could compress it better with more compute time). It's more reasonable to do what Crunchyroll used to do: proactively burn in subs for every language stored as separate video files, but that gives file management woes and only really works when the service is designed to work like that.
If a major streamer can't figure out their caching infrastructure and such to balance the burdens, maybe they need to work on their app so it can handle video overlays on a basic level.
I use Jellyfin rather than Plex, but I know Jellyfin decides whether to direct play or transcode on the server based on the player hardware. This is because many pieces of video playing hardware have very weak general purpose compute and use dedicated hardware for decoding and compositing features. This is the same reason that some formats also trigger transcoding with no overlay (if memory serves Rokus less than 4k don't support HEVC). Ultimately this leaves them designing for lowest common denominator devices and very limited ability to implement interesting features in app. Plex and Jellyfin and such can get away with it in ways that actual services can't because your library being broken is your own fault and problem. I have wondered if it would be possible to implement PGS as a UI-like overlay, but presumably there's a reason no one does.
As for what that means they should do though is messy. To me it means that they should continue to do what they used to of having a copy of the video per language and with burned in subtitles on each. But, as said, it kind of requires that the infrastructure has that very abnormal design. Part of the problem with that design is that you can't as cleanly license the content to be playable on players without that design, such as how historically the Crunchyroll content looked good on Crunchyroll but bad on Amazon. The approach of having it be bad everywhere is a pretty bad solution though. So part of why Sony wants this change is that Crunchyroll has been producing a barely-compatible library for licensing.
So there's a whole mess of conflicting stuff. Crunchyroll should have an app that works in the ideal way, but they also should have a library that works in the ideal way on the Crunchyroll Amazon Prime channel that they don't have control over the code for. They already did effectively the exact thing you were suggesting except that they did it up front rather than on-demand, but it gave them other problems. Specifically they were doing the work twice in different ways for the different distribution channels with some of those channels having objectively superior results.
These are the reasons that my original comment was about using a more flexible standard. Leveraging a more powerful standard is something they could presumably convince other distribution channels to do since it's a net benefit for their apps where "rebuild your infrastructure to work in this weird way ours does" wouldn't realistically ever go anywhere.
Yes, yes, pedantry aside, basic video compositing is something a major service like Crunchyroll needs to help partners get going in return for the non-trivial cut of CR's money they're skimming.
This was a really good read. I know a lot more about subtitling than I did before. I am primarily a dub watcher, but the on screen text subtitles are crucial. It's definitely making me reconsider my subscription.
I'm hoping licenses start going to other services. Just to add to competition to possibly improve subtitling. That or a resurgence of fan subtitling. I know it's not easy, but fans on the internet will churn it out quickly. At least based on my experience following OPM manga
I would not count on competition nor fans to fix this issue. With competition, it's not about being better than your competition, it's about not being worse. If you hold the rights to a show, people will be forced to come to you to watch it, regardless of your quality. So as long as it's not unreadable, people will watch anything sufficiently popular. More than that, I find majority of people don't even care. Often when I talk about anime with people, they talk about things they watched on Netflix. Stuff that I often don't bother with because of their useless subtitling.
One more extreme example is Cyberpunk: Edgerunners. When it aired, my friends were raving about it, but I was sitting around waiting for it to get some fansubs. After a while, I eventually caved and, to no one's surprise, had some major issues in just the first episode. See, at several points, people get phone calls on their AR phones (or whatever they are), and the artists then animate a scrolling transcription of what the other side is saying. So Netflix, in their infinite wisdom, just decided to not put subtitles for that side of the conversation and let us rely on the artists' translation.
However, the animated phone text (if I remember correctly) is some sort of neon green in a really janky font and slightly green background, making it difficult to read at a glance. Worse than that, there were even one or two scenes where it showed the person receiving the call from the front. So the text being a holographic display in front of them, was then mirrored and completely unreadable. As I was watching with the Japanese voice-over and English subtitles, I had to rewind and pause several times just to get the gist of what they were saying. When I vented about this to my friends, they just said, "Did it? Oh, yeah. After your specific example, I guess they did. Well, they do it properly in the rest of the show." They did not.
As for fansubs, I can't help but feel as though manga and anime are just completely different beasts when it comes to translations. Some of my assumptions on it may be a little off as I don't read manga pretty much at all, so keep that in mind. The first, and probably most notable, aspect is that anime already gets same day translations. With manga, I don't believe this is the case at all. So there's an incentive to beat the official translations; you get a lot more prestige, or ad revenue, or whatever it is the translators are chasing by simply staying ahead of the official translation.
The next is the work involved. Anime often covers several chapters of manga in a single episode. Chapters that are released weekly, fortnightly, monthly, or sometimes even less frequently than that. So there's often just more to translate. Granted, with a lot of them, they can rely on the source material that they are likely already familiar with, so they may be ahead of things a tiny bit in a way. But also, the work to make sure things are readable. With manga (and again, this might be my assumptions at play), there's not as much a worry in things being perfectly legible¹ because people can read it at their own pace. With anime, there's very much a limit to what you can do. There's limits on words per minute. When multiple people are talking at the same time, you can't just put each person's text into their respective bubbles, you have to figure out how to convey who's talking. When there's a sign, it needs to be readable at a glance without distracting the viewer away from the things that people are saying. And lastly, the timing; the text needs to line up with certain things to minimize the distraction that the changing subtitles may cause.
And lastly, fansubs have already been dying out. Most the new fansubs I see pop up are things from before Crunchyroll became such a big player. Other times, they're just restylized, upscaled, or fixed versions of a fansub that someone else released. Other times, they're just Crunchyroll/Netflix/whoever's official subs with proper typesetting and signing. I've little doubt the amount of groups that are doing original subtitles for currently airing shows at any given time are countable on one hand, and then they're generally at least a week behind the official subs. Good luck avoiding spoilers if you want to discuss that show.
So, I really don't believe there's any hope that things will improve without Crunchyroll or one of the other major corporations actively deciding to do better. We could theoretically push them towards improving by unsubscribing until they do. But as I said, I don't believe majority of people care enough. So it would be a slow burn and by the time they take measures to improve, the writing for them may already be on the wall.
¹I have no doubt manga translators have some standards and put a lot of work into making sure it's all readable and fits with the art. However, I do believe they occasionally use a less legible font at times in order to convey certain emphasis or tone. Something the original Japanese version is likely doing as well.
You're mostly right. Only some big series like One Piece have simultaneous international releases. Though for those big series, there's a thriving market for acquiring early (usually Chinese) copies then rapidly translating them before the official release.
Yep, it's kind of disappointing, but also the signs are the thing that Netflix et al are most likely to fuck up. Even for dubs, it can be worthwhile to find a good release with translated signs. It's shockingly common for the dub to have no sign translations if a show is less popular...
I've worked as a cleaner for manga fanslations, so I can answer this! It's very very involved if you go all-in on translating everything. The biggest example is sound effects. The Japanese love their sound effects, and it takes a lot of time to carefully remove and replace them. Sadly, I don't have any typeset examples on hand, but I have a few screenshots of a couple panels I cleaned. I was a complete novice when I did these two panels, so you can find some inconsistent gradients and patterns if you zoom in.
Ignoring sound effects (as many official releases do), cleaning is still required for panels where the words are directly written on the art. If they're not in a speech bubble, it takes time and skill to properly erase and replace.
Poor subtitling is actually a big reason I've never considered getting a Crunchyroll or Funimation plan. I'm generally not watching the newest, latest shows, so by the time I'm convinced that something's worth watching, way better fansubs have been released than what you get through the official channels. Sad to say, but unless you're invested in being part of all the online hype every week (or watch anime primarily dubbed, I suppose), I don't see the experience on CR/Amazon/Netflix as better than sailing the high seas.
The irony is that while you used to maybe have to be skeptical about the quality of fansub translations back in the day, now the fansubs are based on, and improve, the official subs.
Even for dubs, translated signs are super useful, and bad typesetting is still distracting. It's somewhat necessary to pirate for older dubs too.